Talk:Low-discrepancy sequence

From Wikipedia, the free encyclopedia

Are these the same as quasi-random or sub-random sequences (that terminology is used in Numerical Recipes)? If so it would be good to say so. Josh Cherry 23:10, 15 May 2004 (UTC)

Hi Josh. Yes, they are the same. I have added quasi-random and sub-random as synomyms, and put NR on the list of references. The NR discussion will probably be more comprehensible to many readers. I've also made quasi-random, sub-random, quasirandom, and subrandom redirect to low-discrepancy sequence in an effort to head of duplication of the topic. Hope this helps, Wile E. Heresiarch 16:17, 16 May 2004 (UTC)

[edit] entry for Jurjen Ferdinand Koksma

Hi,

This page should have a reference to Jurjen Ferdinand Koksma, after whom some of the items on this page are named. He was a professor at the University of Amsterdam.

[edit] Explicit discrepancy calculations

Hello, I wonder if we can put in the discrepancy calculated for some examples. E.g. uniform distribution of points, pseudo-random, maybe some other patterns. I think that those examples would shed some light on low-discrepancy sequences in comparison. Have a good day. 64.48.193.123 18:36, 15 January 2006 (UTC)

[edit] Error in definition?

Is there perhaps an error in the definition given in the first sentence? It says, "for all N, the subsequences x₁, ..., x_N and x₁, ..., x_N+1 are almost uniformly distributed." But that x₁, ..., x_N+1 is almost uniformly distributed follows immediately from the fact thet x₁, ..., x_N is (for all N), when used for N + 1. -- Meni Rosenfeld (talk) 13:12, 15 August 2006 (UTC)

I noticed the same. Further, the intro promises that the notion "almost uniformly distributed" will be made precise, but it is never defined. Presumably the definition is something like "having a low discrepancy", but, although "discrepancy" ("Star-Discrepancy"?) is defined (in an unnecessarily hard-to-understand way; what the h*** is "Niederreiter's notation"?), it is never defined what counts as "low" here. Also, presumably, it is a stronger notion than equidistributed, but no relation is made. --Lambiam^Talk 21:50, 16 August 2006 (UTC)